U.S. PE Firms Scout China's Robot Makers as Automation Threatens Domestic Returns

The competitive intelligence tour has replaced the deal tour. American venture capitalists are traveling to China in growing numbers not to write checks into Chinese companies, but to measure the manufacturing gap that threatens their domestic robotics portfolios. The shift in intent, from capital deployment to competitive reconnaissance, signals a recalculation by US institutional capital about the scale of the challenge facing American robotics startups.

Ryan Cunningham, founder of Edgerunner Ventures, a venture capital firm focused on energy and computing, is organizing an April trip to China for 10 investors. The dynamic has shifted sharply: last year, when Cunningham floated the idea in a WhatsApp group of robotics investors, the reception was tepid. This year, nearly every investor group chat he participates in is circulating notes on recent or upcoming visits to Chinese factories, he told Business Insider .

"The reason I want to bring them to China is because I want to take them by the shoulders and say, 'Guys, if you want to get serious, this is what we should be learning from,'" Cunningham told Business Insider .

The urgency is not abstract. Behind these tours sits a body of data that is reordering assumptions about where the robotics industry's center of gravity actually sits, and what American venture portfolios are genuinely competing against.

The Market Share Numbers That Drove Investors onto Planes

The statistics that are pulling American capital eastward are stark. Chinese companies Unitree and AgiBot together accounted for 71% of all humanoid robots shipped globally last year, according to research firm Omdia . China also controls 63% of the key companies in the global supply chain for humanoid robot components, according to the Mercator Institute for China Studies, a German think tank .

Those two data points, read together, describe a structural condition rather than a temporary lead. Supply chain control means that even American humanoid robot developers depend, in many cases, on Chinese suppliers for components or assembly, according to Stanford's 2026 Emerging Technology Review .

Most US humanoid robots are still in development phases and are not commercially available off the shelf . The commercial deployment gap compounds the data disadvantage: robots in the field generate real-world training data, and robots in development labs do not.

What G2 Venture Partners Found on the Ground

The visits are producing concrete intelligence that is revising conventional assumptions. Neel Mehta, an investor at G2 Venture Partners, a Bay Area firm that spun out of Kleiner Perkins and invests in growth-stage companies building technology for physical industries including energy, manufacturing, and transportation, traveled to Shenzhen and Beijing earlier this year with a colleague .

The trip was anchored around Morgan Stanley's China Summit in Shenzhen, which included field visits to robotics and autonomous-vehicle companies. Mehta and his colleague also arranged visits to Xiaomi and BYD .

Mehta returned with a revised thesis. For years, he said, US investors had operated on the assumption that China might hold a manufacturing edge but that American companies would consistently produce superior AI. He came back convinced both assumptions were wrong .

At a Xiaomi dealership in Shenzhen, Mehta examined the company's SU7 electric vehicle, which starts at approximately $30,000. "The quality and finish is everything you'd expect in a luxury car like Porsche, no exaggeration," Mehta told Business Insider . The vehicle is not sold in the US, where Chinese EV imports face a 100% tariff.

The data infrastructure observation was equally arresting. Mehta saw a data-collection farm where hundreds of workers operated in pods around the clock, teleoperating robots to generate training data at scale . This kind of labor-intensive, high-throughput data generation is structurally difficult for American companies to replicate at comparable cost.

Beyond G2, investors from Eclipse, G2 Venture Partners, and Chemistry VC have made trips to China in recent months. Partners from Founders Fund and Khosla Ventures have also traveled there, according to a report from The Information .

Policy Risk Layers onto Competitive Risk

The competitive pressure exists inside an accelerating regulatory envelope. In July 2026, the Federal Communications Commission banned new foreign-made "advanced robotic devices," citing what it described as unacceptable risks to national security . The ruling reflects a broader US government posture that treats robotics as a national-security sector rather than a purely commercial one.

China's government has been equally deliberate. In a 2023 policy document, Beijing drew an explicit comparison between the potential economic impact of humanoid robots and the historical impact of computers, smartphones, and electric vehicles, and set a target to deploy humanoid robots across the broader economy by 2027 .

"The Chinese have been very hardware focused," Wendy Chang, a senior analyst at the Mercator Institute for China Studies, told Business Insider. "They are trying to own the whole robotics supply chain."

The hardware-first strategy diverges from Silicon Valley's historical software-and-platform playbook. That divergence is precisely what makes the competitive assessment so difficult for US investors to perform from home.

The technology competition extends beyond robotics hardware. General Motors and Ford are deepening software and autonomous-driving strategies with support from Google and Nvidia, according to Digitimes . The parallel dynamic, American legacy companies turning to Silicon Valley to close a technology gap just as Silicon Valley turns to China to understand a manufacturing gap, captures the layered nature of the competitive challenge in physical AI.

The Plocamium View

Our view: the factory tours are a leading indicator of a portfolio construction shift, not just a learning exercise. When venture investors of the caliber of Founders Fund and Khosla Ventures begin making competitive intelligence trips in person, the implication is that their existing portfolio assumptions are under review.

The 71% shipment share held by Unitree and AgiBot is not a near-term problem for US robotics startups, it is a compounding one. Deployed robots generate training data. Training data improves model performance. Better models reduce the cost and failure rate of further deployments. Each cycle widens the lead. American humanoid companies still in development are not generating that data loop. They are, in effect, starting a race that Chinese competitors began running years ago.

The supply chain figure from the Mercator Institute deserves equal weight. A 63% control position across key humanoid component suppliers is not easily disrupted by tariffs or FCC bans. It is built from years of industrial policy, capital allocation, and manufacturing scale. The July FCC ban on foreign-made advanced robotic devices addresses the demand side of that equation. It does nothing to accelerate the US supply-side capability.

The second-order play for institutional capital is not in the humanoid robot makers themselves, at least not at current valuations. The more defensible investment thesis sits in the enabling layer: the software, simulation, and data infrastructure that allows US robotics companies to close the training-data gap without replicating China's labor-cost advantage. Companies building synthetic data generation, digital twin environments, and teleoperation platforms for robot training are addressing exactly the bottleneck that Mehta observed in Shenzhen.

The FCC ban and the broader national-security framing of robotics also create a durable moat for US-domiciled supply chain alternatives. The reshoring premium, meaning the price gap between domestically sourced components and Chinese-sourced equivalents, will narrow as policy pressure sustains capital allocation toward domestic production. PE firms with manufacturing sector expertise are better positioned to capture that transition than pure-play venture capital.

China's 2027 deployment target, set in the 2023 policy document, is now less than two years away. Whether Beijing hits that target precisely is less important than the capital and industrial infrastructure being assembled to pursue it. American investors touring factories in Shanghai are not tourists. They are performing the kind of ground-level due diligence that should have begun earlier. That they are doing it now, in organized delegations, with translators and five-star hotels included in the price of admission, suggests the reckoning with China's robotics lead has moved from denial to analysis.

The next move is capital reallocation. Watch for it.

The Bottom Line

China holds a structural lead in humanoid robotics: 71% of global shipments, 63% of the supply chain, and a government-mandated deployment target for 2027. American venture capital has spent years discounting Chinese hardware quality and AI capability. The investors now boarding planes to Shanghai are correcting that error in real time. For institutional portfolios with robotics exposure, the question is no longer whether China is ahead. The question is which US companies have a credible plan to compete anyway, and which are simply hoping tariffs hold the gap open long enough for them to catch up.